Much excitement has been created this week by an analysis of YouGov polling data released by the anti-Brexit Best for Britain campaign and first reported by The Observer. Using a statistical technique (multi-level regression and post-stratification) that, inter alia, helped YouGov anticipate that the Conservatives would lose their overall majority in last year’s general election, the analysis identified 112 constituencies where it had previously been estimated by Prof. Chris Hanretty that a majority of voters had voted to Leave in 2016 but where a majority were now estimated to be in favour of Remain. Described by Best for Britain as ‘a monumental shift in public opinion’, the impression, at least, was given that the analysis was evidence of a significant new swing of public opinion in favour of Remain.
The reality, however, was much more prosaic. The polling data on which the analysis was based is primarily, though not exclusively, a large 10,000 sample poll conducted earlier this month by YouGov for another anti-Brexit campaign, The People’s Vote, which is campaigning for a referendum on whatever Brexit deal eventually emerges. The data were reported as showing that support for Remain across the country as a whole now stands at 53%, while support for Leave was estimated to be 47% (which, indeed, were the figures that had been previously been published for the People’s Vote poll in particular). That, of course, represents a five-point swing from Leave to Remain as compared with the result of the 2016 referendum, when Leave won 52% and Remain 48%. But it is more or less in line with other polling conducted during the last three months which, on average has put Remain on 52%, Leave 48%. So there is no evidence in these numbers of a significant new swing to Remain. Rather it simply represents confirmation of other recent polling that Remain appear to be slightly ahead.
However, that still leaves the estimate of 112 constituencies where in 2016 there was a Leave majority but where Remain are now thought to be ahead. That sounds like a lot of constituencies. Indeed, it is sufficient to ensure that, in contrast to the position in 2016, over half (341) of the 632 constituencies in Britain are now thought to contain a majority of Remain supporters. However, the figure is but a reflection of the relatively even spread of the Leave vote across much of provincial England together with Wales in the 2016 referendum. There were no less than 115 constituencies in 2016 where Prof. Hanretty estimated that Leave won between 50% and 55% of the vote. If the 5% swing since 2016 implied by YouGov’s polling had occurred in each and every constituency, that would be sufficient to turn all of these 115 seats from being majority Leave seats to majority Remain constituencies – albeit only narrowly. In other words, it would be surprising if any analysis of polling based on a 53% Remain vote across the country as a whole did anything other than identify over 100 constituencies where the majority outcome would now be different.
True, the analysis that underlay Best for Britain’s headline figure did much more than simply assume the whole country had swung to the same extent. The statistical technique that was employed uses polling data to identify the probability that voters with particular characteristics will vote Remain or Leave, examines government statistics and other sources to calculate the proportion of people with those characteristics in each constituency, and then combines the two to estimate the likely outcome of a ballot in each seat. Consequently, if particular kinds of voters (perhaps in particular kinds of places) have swung from Leave to Remain, then the analysis should find a bigger swing to Remain in places with more such voters – and conversely a lower swing elsewhere. But the fact that the analysis emerged with more or less the same number of seats swinging from Leave to Remain as would be anticipated from a uniform movement across the country as a whole means that the estimated variation in the swing had little or no net impact on the total number of seats whose status is thought to have changed.
A Shift Within Labour’s Ranks?
However, that does not mean that the variation which is said to have been identified is not of interest. One of the aims of those currently campaigning against Brexit appears to be to try to persuade the Labour party in particular to change its stance, and at least come out in favour of a second referendum if not indeed to oppose Brexit entirely. Thus, it was notable that the Best for Britain analysis is reported as showing that the swing from Leave to Remain ‘has been driven by doubts among Labour voters who backed Leave’ and that, consequently, it is greatest in Labour heartlands in the North of England and Wales. Doubtless anti-Brexit campaigners are hoping that this finding will help persuade Labour MPs representing constituencies where a majority voted Leave in 2016 that a change in the party’s stance would not be so harmful electorally as some of them at least seem to fear, albeit that we have previously shown that, even outside London, a majority of Labour voters in Labour seats voted for Remain. This finding was certainly quoted eagerly this week by the former Labour Prime Minister, Gordon Brown.
But what evidence is there in the polls that Labour voters who backed Leave are now particularly likely to be having second thoughts? A look at the data from the large poll conducted for the People’s Vote campaign certainly suggests that Labour’s vote now looks to be even more pro-Remain than it was at the time of the 2017 election (when the party was more successful at winning over Remain than Leave voters). The poll suggests that no less than 77% of those who now say that they would vote Labour would now also vote Remain. In contrast, at the time of the 2017 election YouGov estimated that 68% of those who voted Labour then had voted Remain in 2016. (Meanwhile, at 74%, the proportion of Conservative voters that favour Leave still looks to be much the same as it was in June of last year.)
There is, though, more than one possible reason why the balance of opinion might have tilted yet further in favour of Remain amongst Labour supporters. One is that Labour have been more successful in retaining and/or winning over the support of those who back Remain than it has those who supported Leave. Of this there is some sign. While, as already noted, 77% of those who currently support Labour are also currently Remain supporters, the equivalent figure (in the same YouGov/People’s Vote poll) for those who voted Labour a year ago is, at 75%, slightly lower. However, while this difference might mean that Labour have gained more Remain than Leave supporters in the last 12 months, it could also mean that the party has lost more ground among those who back Leave than those who favour Remain. This is something that ICM in particular have been tracking. And on average in their last four polls, only 77% of 2017 Labour voters who backed Leave in 2016 now say they would vote Labour again, compared with 84% of those who supported Remain.
Still, this is evidently not a sufficient explanation for the higher level of support for Remain amongst Labour voters identified by YouGov. What evidence is there, then, that Labour voters have shifted from Leave to Remain? The fact that 75% of YouGov’s 2017 Labour voters now say that they back Remain, whereas in 2016 only 68% of them did so, certainly implies (though does not prove) that this indeed may have happened. But if this were a sign of a recent conversion from Leave to Remain amongst Labour voters we would expect to find that those who voted Labour in 2017 are keener on Remain now than they were at the time of last year’s election. However, according to data collected by Survation at least, that is not what has happened. In three polls that that company conducted immediately after the 2017 election they found on average that at that time, 72% of those who voted Labour backed Remain. But in three polls that Survation conducted in June and July this year, the company found that, at 69%, the proportion of 2017 Labour voters that now support Remain is fact not higher, but slightly lower.
So how do we make sense of this seemingly contradictory evidence? Well, there is another group of citizens who are, perhaps, all too easily forgotten in the current febrile political atmosphere – those who did not vote in the EU referendum and, indeed, those who did not vote in the 2017 general election. This group’s attitude to Brexit and their party choice is decidedly distinctive.
It has been evident for some time that those who did not vote in 2016 are more likely to say that they would now vote Remain rather than Leave. That this is still the case has been confirmed by recent polling by Deltapoll; in two polls the company conducted between May and July, 44% of those who did not vote in 2016 said that they would now back Remain, while only 16% indicated support for Leave. Much of the overall narrow lead that Remain currently enjoys in the polls is a product of this imbalance amongst non-voters.
But Deltapoll’s analyses add a new piece to the jigsaw. They indicate that those who did not vote in 2016 are not only more likely to back Remain, but also are much more likely to support Labour. The company’s recent polls suggest that around three-fifths back Labour, while no more than one in five, and maybe even fewer, would vote Conservative. Indeed, these polls also have Labour well ahead amongst those who did not vote in the 2017 election, suggesting that some of these new supporters have come to support Labour within the last twelve months.
Part of the explanation for this pattern, of course, will be that some of these new voters will have turned 18 within the last two years, and given that young voters were much more likely to vote Remain in 2016 and Labour in 2017, we might well expect this cohort of new voters to help swell the ranks of Remain supporters within Labour’s ranks. In any event, Deltapoll’s evidence is a reminder that the battle for public opinion over Brexit is not simply a battle between the existing serried ranks of Remain and Leave supporters. It is also a contest for the eyes and ears of those who did not vote two years ago. And in Labour’s case in particular it looks as though as it could be a contest that matters as it considers the electoral politics of Brexit.
A shorter version of this blog appears on The Conversation website.